> I'd put off doing it. We lost our cell entirely and had to spend a lot
> of time with Transarc on the phone trying to get back up. These were
> Solaris servers, but I don't think it really matters what platform you're
> on. We're still running with 3.4 everything except for the VL server.
I think this whole thing has been blown out of proportion. Yes,
it's a bad bug, and it certainly should be fixed. On the other
hand, rebuilding a VLDB from scratch is certainly not a difficult
task. About 6 months ago, we (finally) upgraded our database servers
from 3.2 + CMU mods to (almost stock) 3.3a. We didn't use the VLDB
conversion programs, and never had any intention of doing so. We
had planned to rebuild the VLDB from scratch, and that's exactly
what we did. For our cell, which at the time had about 40 fileservers
and 250GB online, that rebuild took no more than a couple of hours,
starting with an empty database. It's a simple matter of running
vos syncvldb against each server in turn; I don't recall that any
other step was necessary.
Note that in our case, we reduced the cutover time by building the
new database servers on separate machines. This was necessary because
we were switching operating systems at the same time, but can be
accomplished almost as easily by temporarily setting up an extra
machine (even someone's office machine) to think it's the VLDB server,
and running the syncvldb operations there.
-- Jeffrey T. Hutzelman (N3NHS) <[EMAIL PROTECTED]>
Systems Programmer, CMU SCS Research Facility
Please send requests and problem reports to [EMAIL PROTECTED]
